9697

• UCREFRP Catalog Number: 9697
• Author: Bestgen, K.R., D.W. Beyers, J.A. Rice and G.B. Haines.
• Title: Factors affecting recruitment of young Colorado pikeminnow
• USFW Year: 2006.
• USFW - Doc Type: synthesis of predation experiments, field studies, and individual-based modeling.
• Copyright Material: YES

1728

Colorado pikeminnow larvae (age-O) and age-1 to
juvenile fish (hereafter collectively called juveniles)
were captured from backwaters during summer Oate
Iuly to early August) and autumn (late September to
early October). Colorado pikeminnow were taken with
a seine (4.6 X 1.3 m; 1.6-mm mesh), preserved in
100% solutions of ethanol, and retained for otolith
analyses.
Specimen analysis.--Colorado pikeminnow larvae
from drift-net collections (<20 mm TL, but 99% were
<10 mm TL) and fish captured in seine collections in
summer (9.0--25.4-mm TL) and autumn (15.9--66.9-
mm TL) were measured to the nearest O.l-mm via an
ocular micrometer or calipers. Ages were estimated by
counting daily increments in the left sagitta Oarvae) or
lapillus (larvae and juveniles) and averaging two to
four independent counts. Otoliths from all but 9 of
1,929 fish exanained were readable. Hatching date for
each fish was calculated by subtracting its estimated
age (d) from its date of capture. Regression analysis
showed that ages of Colorado pikeminnow reared in
water with daily temperature fluctuations, such as
occur in backwaters of the Green River, can be
accurately estimated up to 165 d after hatching using
this method (Bestgen and Bundy 1998).
Hatching date distributions.-Distributions of
hatching dates were used to assign fish to one of three
or four intra-annual cohorts. Cohorts were divided to
represent early, middle, and late summer reproductive
output, and there was often a distribution mode
associated with each cohort. The number and distribu-
tion of larvae and juveniles in each cohort were
compared to determine if observed and expected
survival were similar. In this analysis, expected
distributions represented larvae captured in drift-net
samples. Observed distributions were autumn juveniles
from the same cohorts. We expected that the relative
proportion of autumn juveniles captured in seine
samples (observed distributions) should be similar to
the proportion of larvae in expected distributions in the
absence of size-dependent mortality. To account for
sample-size differences for larvae and jUveniles, the
expected number of juveniles in each cohort was
estimated by calculating the proportion of larvae in
each cohort and multiplying this value by the total
number of juveniles in each autumn sample. A log-
likelihood ratio test (G-test; Sokal and Rohlf 1969) was
used to compare observed and expected proportions of
autumn juveniles in cohorts.
One factor that may bias the results of this analysis is
that wild larvae may suffer differential mortality
through the summer because of differences in hatching
time, particularly if sources of mortality other than that
from red shiner predation exist (e.g., birds, larger

BESTGEN ET AL.

piscivores). This may be true because early hatched
larvae are at large for longer periods and, therefore,
may undergo higher mortality than larvae hatched later,
even if they have outgrown susceptibility to red shiner
predation. We evaluated the importance of this
potential bias by imposing a constant mortality rate
of 5% per day on larvae so that early hatched fish
would suffer higher cumulative mortality through the
summer than late hatched fish (Bestgen 1997). The 5%
per day mortality rate was selected because it resulted.
in a survival rate for a group of larvae over a 9O-d
period of less than 1 %, a level which seemed
reasonable based on studies conducted on species with
similar-sized larvae (Houde 1987). The results of that
analysis suggested fewer fish from early hatched
cohorts remained in autumn than in the absence of
the mortality factor, but conclusions about the rank
order of relative survival rates and the contribution of
juvenile cohorts to annual recruitment were unchanged.
Thus, our conclusions about the relative contribution of
various cohorts of larvae to the population of juveniles
surviving to autumn were probably not biased by the
amount of time larvae were at large.
A survival index of cohorts within a year (observed
proportion of juveniles in a cohort/expected propor-
tion) were compared to evaluate the relative contribu-
tions of various cohorts of larvae to recruitment
(Crecco and Savoy 1985). Survival index values of
1.0 indicated that juveniles from a cohort occurred in
equal proportion to larvae, whereas greater values
indicated relative survival greater than that expected
and lesser values less than that expected.
Growth rates.-Growth rates (mm/d) of juvenile
Colorado pikeminnow captured in backwaters of the
Green River were calculated by subtracting 5.5 mm
(mean TL at hatch; Bestgen and Williams 1994) from
TL at capture and dividing by age (d), which was
estimated from otoliths. The mean and distribution of
growth rates of juveniles within a cohort were
qualitatively compared between summer and autumn
samples to evaluate whether size-selective mortality
was occurring between the two sampling dates. Growth
rates of Colorado pikeminnow juveniles in summer that
averaged less than those of fish that survived to autumn
would indicate that relatively small and slow-growing
fish were being selectively removed.

Components of the Individual-Based Simulation Model

Number of attacks.-Equation (3) was used to
estimate the number of predator attacks per day on
individual larvae based on environmental conditions.
For each larva on each day, equation (3) was solved
using larva TL, water clarity condition, alternative prey
condition, and one of the five feeding strategies